Fix regression described in:
https://github.com/git-for-windows/git/issues/1392
which was introduced in:
b2353379bb
Problem Symptoms
================
When the user has a .gitignore file that is a symlink, the fscache
optimization introduced above caused the stat-data from the symlink,
rather that of the target file, to be returned. Later when the ignore
file was read, the buffer length did not match the stat.st_size field
and we called die("cannot use <path> as an exclude file")
Optimization Rationale
======================
The above optimization calls lstat() before open() primarily to ask
fscache if the file exists. It gets the current stat-data as a side
effect essentially for free (since we already have it in memory).
If the file does not exist, it does not need to call open(). And
since very few directories have .gitignore files, we can greatly
reduce time spent in the filesystem.
Discussion of Fix
=================
The above optimization calls lstat() rather than stat() because the
fscache only intercepts lstat() calls. Calls to stat() stay directed
to the mingw_stat() completly bypassing fscache. Furthermore, calls
to mingw_stat() always call {open, fstat, close} so that symlinks are
properly dereferenced, which adds *additional* open/close calls on top
of what the original code in dir.c is doing.
Since the problem only manifests for symlinks, we add code to overwrite
the stat-data when the path is a symlink. This preserves the effect of
the performance gains provided by the fscache in the normal case.
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Make fscache_enabled() function public rather than static.
Remove unneeded fscache_is_enabled() function.
Change is_fscache_enabled() macro to call fscache_enabled().
is_fscache_enabled() now takes a pathname so that the answer
is more precise and mean "is fscache enabled for this pathname",
since fscache only stores repo-relative paths and not absolute
paths, we can avoid attempting lookups for absolute paths.
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Teach read_directory_recursive() and add_excludes() to
be aware of optional fscache and avoid trying to open()
and fstat() non-existant ".gitignore" files in every
directory in the worktree.
The current code in add_excludes() calls open() and then
fstat() for a ".gitignore" file in each directory present
in the worktree. Change that when fscache is enabled to
call lstat() first and if present, call open().
This seems backwards because both lstat needs to do more
work than fstat. But when fscache is enabled, fscache will
already know if the .gitignore file exists and can completely
avoid the IO calls. This works because of the lstat diversion
to mingw_lstat when fscache is enabled.
This reduced status times on a 350K file enlistment of the
Windows repo on a NVMe SSD by 0.25 seconds.
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Teach "add" to use preload-index and fscache features
to improve performance on very large repositories.
During an "add", a call is made to run_diff_files()
which calls check_remove() for each index-entry. This
calls lstat(). On Windows, the fscache code intercepts
the lstat() calls and builds a private cache using the
FindFirst/FindNext routines, which are much faster.
Somewhat independent of this, is the preload-index code
which distributes some of the start-up costs across
multiple threads.
We need to keep the call to read_cache() before parsing the
pathspecs (and hence cannot use the pathspecs to limit any preload)
because parse_pathspec() is using the index to determine whether a
pathspec is, in fact, in a submodule. If we would not read the index
first, parse_pathspec() would not error out on a path that is inside
a submodule, and t7400-submodule-basic.sh would fail with
not ok 47 - do not add files from a submodule
We still want the nice preload performance boost, though, so we simply
call read_cache_preload(&pathspecs) after parsing the pathspecs.
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Teach FSCACHE to remember "not found" directories.
This is a performance optimization.
FSCACHE is a performance optimization available for Windows. It
intercepts Posix-style lstat() calls into an in-memory directory
using FindFirst/FindNext. It improves performance on Windows by
catching the first lstat() call in a directory, using FindFirst/
FindNext to read the list of files (and attribute data) for the
entire directory into the cache, and short-cut subsequent lstat()
calls in the same directory. This gives a major performance
boost on Windows.
However, it does not remember "not found" directories. When STATUS
runs and there are missing directories, the lstat() interception
fails to find the parent directory and simply return ENOENT for the
file -- it does not remember that the FindFirst on the directory
failed. Thus subsequent lstat() calls in the same directory, each
re-attempt the FindFirst. This completely defeats any performance
gains.
This can be seen by doing a sparse-checkout on a large repo and
then doing a read-tree to reset the skip-worktree bits and then
running status.
This change reduced status times for my very large repo by 60%.
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
If multiple threads access a directory that is not yet in the cache, the
directory will be loaded by each thread. Only one of the results is added
to the cache, all others are leaked. This wastes performance and memory.
On cache miss, add a future object to the cache to indicate that the
directory is currently being loaded. Subsequent threads register themselves
with the future object and wait. When the first thread has loaded the
directory, it replaces the future object with the result and notifies
waiting threads.
Signed-off-by: Karsten Blees <blees@dcon.de>
Checking the work tree status is quite slow on Windows, due to slow
`lstat()` emulation (git calls `lstat()` once for each file in the
index). Windows operating system APIs seem to be much better at scanning
the status of entire directories than checking single files.
Add an `lstat()` implementation that uses a cache for lstat data. Cache
misses read the entire parent directory and add it to the cache.
Subsequent `lstat()` calls for the same directory are served directly
from the cache.
Also implement `opendir()`/`readdir()`/`closedir()` so that they create
and use directory listings in the cache.
The cache doesn't track file system changes and doesn't plug into any
modifying file APIs, so it has to be explicitly enabled for git functions
that don't modify the working copy.
Note: in an earlier version of this patch, the cache was always active and
tracked file system changes via ReadDirectoryChangesW. However, this was
much more complex and had negative impact on the performance of modifying
git commands such as 'git checkout'.
Signed-off-by: Karsten Blees <blees@dcon.de>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Add a macro to mark code sections that only read from the file system,
along with a config option and documentation.
This facilitates implementation of relatively simple file system level
caches without the need to synchronize with the file system.
Enable read-only sections for 'git status' and preload_index.
Signed-off-by: Karsten Blees <blees@dcon.de>
Emulating the POSIX lstat API on Windows via GetFileAttributes[Ex] is quite
slow. Windows operating system APIs seem to be much better at scanning the
status of entire directories than checking single files. A caching
implementation may improve performance by bulk-reading entire directories
or reusing data obtained via opendir / readdir.
Make the lstat implementation pluggable so that it can be switched at
runtime, e.g. based on a config option.
Signed-off-by: Karsten Blees <blees@dcon.de>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Emulating the POSIX `dirent` API on Windows via
`FindFirstFile()`/`FindNextFile()` is pretty staightforward, however,
most of the information provided in the `WIN32_FIND_DATA` structure is
thrown away in the process. A more sophisticated implementation may
cache this data, e.g. for later reuse in calls to `lstat()`.
Make the `dirent` implementation pluggable so that it can be switched at
runtime, e.g. based on a config option.
Define a base DIR structure with pointers to `readdir()`/`closedir()`
that match the `opendir()` implementation (similar to vtable pointers in
Object-Oriented Programming). Define `readdir()`/`closedir()` so that
they call the function pointers in the `DIR` structure. This allows to
choose the `opendir()` implementation on a call-by-call basis.
Make the fixed-size `dirent.d_name` buffer a flex array, as `d_name` may
be implementation specific (e.g. a caching implementation may allocate a
`struct dirent` with _just_ the size needed to hold the `d_name` in
question).
Signed-off-by: Karsten Blees <blees@dcon.de>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
We will use them in the upcoming "FSCache" patches (to accelerate
sequential lstat() calls).
Signed-off-by: Karsten Blees <blees@dcon.de>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
This is the branch thicket of patches in Git for Windows that are
considered ready for upstream. To keep them in a ready-to-submit shape,
they are kept as close to the beginning of the branch thicket as
possible.
To verify that the `clean` side of the `clean`/`smudge` filter code is
correct with regards to LLP64 (read: to ensure that `size_t` is used
instead of `unsigned long`), here is a test case using a trivial filter,
specifically _not_ writing anything to the object store to limit the
scope of the test case.
As in previous commits, the `big` file from previous test cases is
reused if available, to save setup time, otherwise re-generated.
Signed-off-by: Philip Oakley <philipoakley@iee.email>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
To complement the `--stdin` and `--literally` test cases that verify
that we can hash files larger than 4GB on 64-bit platforms using the
LLP64 data model, here is a test case that exercises `hash-object`
_without_ any options.
Just as before, we use the `big` file from the previous test case if it
exists to save on setup time, otherwise generate it.
Signed-off-by: Philip Oakley <philipoakley@iee.email>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Just like the `hash-object --literally` code path, the `--stdin` code
path also needs to use `size_t` instead of `unsigned long` to represent
memory sizes, otherwise it would cause problems on platforms using the
LLP64 data model (such as Windows).
To limit the scope of the test case, the object is explicitly not
written to the object store, nor are any filters applied.
The `big` file from the previous test case is reused to save setup time;
To avoid relying on that side effect, it is generated if it does not
exist (e.g. when running via `sh t1007-*.sh --long --run=1,41`).
Signed-off-by: Philip Oakley <philipoakley@iee.email>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Continue walking the code path for the >4GB `hash-object --literally`
test to the hash algorithm step for LLP64 systems.
This patch lets the SHA1DC code use `size_t`, making it compatible with
LLP64 data models (as used e.g. by Windows).
The interested reader of this patch will note that we adjust the
signature of the `git_SHA1DCUpdate()` function without updating _any_
call site. This certainly puzzled at least one reviewer already, so here
is an explanation:
This function is never called directly, but always via the macro
`platform_SHA1_Update`, which is usually called via the macro
`git_SHA1_Update`. However, we never call `git_SHA1_Update()` directly
in `struct git_hash_algo`. Instead, we call `git_hash_sha1_update()`,
which is defined thusly:
static void git_hash_sha1_update(git_hash_ctx *ctx,
const void *data, size_t len)
{
git_SHA1_Update(&ctx->sha1, data, len);
}
i.e. it contains an implicit downcast from `size_t` to `unsigned long`
(before this here patch). With this patch, there is no downcast anymore.
With this patch, finally, the t1007-hash-object.sh "files over 4GB hash
literally" test case is fixed.
Signed-off-by: Philip Oakley <philipoakley@iee.email>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Continue walking the code path for the >4GB `hash-object --literally`
test. The `hash_object_file_literally()` function internally uses both
`hash_object_file()` and `write_object_file_prepare()`. Both function
signatures use `unsigned long` rather than `size_t` for the mem buffer
sizes. Use `size_t` instead, for LLP64 compatibility.
While at it, convert those function's object's header buffer length to
`size_t` for consistency. The value is already upcast to `uintmax_t` for
print format compatibility.
Note: The hash-object test still does not pass. A subsequent commit
continues to walk the call tree's lower level hash functions to identify
further fixes.
Signed-off-by: Philip Oakley <philipoakley@iee.email>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
The previous commit adds a test that demonstrates a problem in the
`hash-object --literally` command, manifesting in an unnecessary file
size limit on systems using the LLP64 data model (which includes
Windows).
Walking the affected code path is `cmd_hash_object()` >> `hash_fd()` >>
`hash_literally()` >> `hash_object_file_literally()`.
The function `hash_object_file_literally()` is the first with a file
length parameter (via a mem buffer). This commit changes the type of
that parameter to the LLP64 compatible `size_t` type.
There are no other uses of the function. The `strbuf` type is already
`size_t` compatible.
Note: The hash-object test does not yet pass. Subsequent commits will
continue to walk the call tree's lower level functions to identify
further fixes.
Signed-off-by: Philip Oakley <philipoakley@iee.email>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
On LLP64 systems, such as Windows, the size of `long`, `int`, etc. is
only 32 bits (for backward compatibility). Git's use of `unsigned long`
for file memory sizes in many places, rather than size_t, limits the
handling of large files on LLP64 systems (commonly given as `>4GB`).
Provide a minimum test for handling a >4GB file. The `hash-object`
command, with the `--literally` and without `-w` option avoids
writing the object, either loose or packed. This avoids the code paths
hitting the `bigFileThreshold` config test code, the zlib code, and the
pack code.
Subsequent patches will walk the test's call chain, converting types to
`size_t` (which is larger in LLP64 data models) where appropriate.
Signed-off-by: Philip Oakley <philipoakley@iee.email>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Ensure key CMake option values are part of the CMake output to
facilitate user support when tool updates impact the wider CMake
actions, particularly ongoing 'improvements' in Visual Studio.
These CMake displays perform the same function as the build-options.txt
provided in the main Git for Windows. CMake is already chatty.
The setting of CMAKE_EXPORT_COMPILE_COMMANDS is also reported.
Include the environment's CMAKE_EXPORT_COMPILE_COMMANDS value which
may have been propogated to CMake's internal value.
Testing the CMAKE_EXPORT_COMPILE_COMMANDS processing can be difficult
in the Visual Studio environment, as it may be cached in many places.
The 'environment' may include the OS, the user shell, CMake's
own environment, along with the Visual Studio presets and caches.
See previous commit for arefacts that need removing for a clean test.
Signed-off-by: Philip Oakley <philipoakley@iee.email>
In Git-for-Windows, work on using ARM64 has progressed. The
commit 2d94b77b27 (cmake: allow building for Windows/ARM64, 2020-12-04)
failed to notice that /compat/vcbuild/vcpkg_install.bat will default to
using the "x64-windows" architecture for the vcpkg installation if not set,
but CMake is not told of this default. Commit 635b6d99b3 (vcbuild: install
ARM64 dependencies when building ARM64 binaries, 2020-01-31) later updated
vcpkg_install.bat to accept an arch (%1) parameter, but retained the default.
This default is neccessary for the use case where the project directory is
opened directly in Visual Studio, which will find and build a CMakeLists.txt
file without any parameters, thus expecting use of the default setting.
Also Visual studio will generate internal .sln solution and .vcxproj project
files needed for some extension tools. Inform users of the additional
.sln/.vcxproj generation.
** How to test:
rm -rf '.vs' # remove old visual studio settings
rm -rf 'compat/vcbuild/vcpkg' # remove any vcpkg downloads
rm -rf 'contrib/buildsystems/out' # remove builds & CMake artifacts
with a fresh Visual Studio Community Edition, File>>Open>>(git *folder*)
to load the project (which will take some time!).
check for successful compilation.
The implicit .sln (etc.) are in the hidden .vs directory created by
Visual Studio.
Signed-off-by: Philip Oakley <philipoakley@iee.email>
The CMakeSettings.json file is tool generated. Developers may track it
should they provide additional settings.
Signed-off-by: Philip Oakley <philipoakley@iee.email>
As mentioned in the Makefile and CMakeLists.txt: "When cross-compiling, define HOST_CPU as the canonical name of the CPU on which the built Git will run (for instance "x86_64")"
This commit sets the HOST_CPU variable since Git for Windows arm64 is cross-compiled from an amd64 host.
Signed-off-by: Dennis Ameling <dennis@dennisameling.com>
This is needed for the next change, where we add HOST_CPU support to the
CMake definition.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
This simplifies the workflow dramatically.
Note that we have to reinstate that `/usr/bin/git` hack (a shell script
that simply redirects to `/mingw64/bin/git.exe`) in the `pkg` job
manually, since we no longer cache the `build-installers` artifact
_after_ installing that hack in `bundle-artifacts`.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
This merges the current version of the patch that tries to address Git
GUI's problems with intent-to-add files.
This patch will likely be improved substantially before it is merged
into Git GUI's main branch, but we want to have _something_ resembling a
fix already in Git for Windows v2.29.0.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
